Bank of America fined $225 million for faulty automated fraud filter on unemployment cards
Federal regulators fined Bank of America $225 million for botching the disbursement of state unemployment benefits at the height of the pandemic. The bank’s faulty automated fraud detection program allegedly froze legitimate accounts, denying some beneficiaries access to funds.
Bank of America used a faulty fraud detection program that automatically froze legitimate cardholders' accounts.
Key facts
- What
- Federal regulators fined Bank of America $225 million for botching the disbursement of state unemployment benefits at the height of the pandemic.
- Incident date
- Jan 1, 2020
- Who
- Bank of America
- Failure mode
- Hallucination
- AI surface
- Search / RAG
- Severity
- High
What happened
Bank of America was contracted by state agencies to distribute unemployment benefits via prepaid debit cards. The bank deployed an automated fraud filter that incorrectly flagged and froze legitimate cardholders’ accounts. This resulted in many vulnerable consumers being denied access to mandated funds, with limited means to remedy the situation.
What broke inside the model
- 01 · TriggerA user asks for a fact, a citation, or a figure.
- 02 · Model stepThe model writes a fluent, confident answer.
- 03 · Control gapNothing ties the claim back to a real source.
- 04 · FailureA fabricated fact ships as if it were verified.
- 05 · ConsequenceThe false claim reaches a customer, a court, or the public.
Confidence holds, and even spikes, as the claim detaches from any source.
The automated fraud detection system produced excessive false positives, triggering widespread account freezes for legitimate users. The system lacked a prompt and adequate investigation process to resolve these errors, leading to prolonged loss of funds.
What it cost
Sources
Cite this entry
https://failureindex.ai/failures/bank-america-fined-225-million-faultyAI Failure Index. "Bank of America fined $225 million for faulty automated fraud filter on unemployment cards" (FI-0220). Realm Labs. https://failureindex.ai/failures/bank-america-fined-225-million-faulty (indexed Jun 5, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0220. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
A runtime layer that watches the model's internal state can flag the moment a model commits to a claim it has no support for, and hold or reroute the response before it reaches a user. Realm reads those signals in real time rather than grading the transcript after the fact.